Structure of the lipopolysaccharide from an Escherichia coli heptose-less mutant. I. Chemical degradations and identification of products.

The structure of lipopolysaccharide from a heptose-less mutant of Escherichia coli K-12 has been investigated. Lipopolysaccharide isolated from 32P-labeled cells was treated with mild alkali to yield two separable components: [OH-LPS]-I (approximately 70%) and [OH-LPS]-II (approximately 30%). Mild acidic treatment of [OH-LPS]-I gave mainly a product which was identified as (4-O-phosphoryl-N-beta-hydroxymyristyl-D-glucosaminyl)-beta(1 leads to 6)-N-beta-hydroxymyristyl-D-glucosamine 1-phosphate (Compound I). Further acidic hydrolysis of both [OH-LPS]-I and [OH-LPS]-II yielded as the main product (4-O-phosphoryl-N-beta-hydroxymyristyl-D-glucosaminyl)-beta(1 leads to 6)-N-beta-hydroxymyristyl-D-glucosamine (Compound II). The structures of the above products were deduced by a combination of compositional analyses, sensitivity to phosphomonoesterase, rates of hydrolysis of the phosphate groups and alkali-catalyzed beta elimination of the phosphate residues following appropriate oxidation of hydroxyl groups. These studies together with work reported in the accompanying papers have led to the identification of two species of lipopolysaccharide in the E. coli strain both of which contain a single glucosamine dissacharide unit but differ in having monosubstituted phosphate or pyrophosphate groups at the glycosidic position. Each species of lipopolysaccharide also appeared to be heterogeneous with respect to the number of esterified fatty acyl groups.

The structure of lipopolysaccharide from a heptoseless mutant of Escherichia coli K-12 has been investigated.
Lipopolysaccharide isolated from 32P-labeled cells was treated with mild alkali to yield two separable components: [  to consist of three distinct regions (1,2). The outermost region consists of a polysaccharide chain with repeating oligosaccharide units of variable length and constitutes the O-antigen region. The latter is connected to the core polysaccharide chain, which is relatively invariant within certain classes of bacteria. The core polysaccharide is linked through one or more units of a unique g-carbon sugar, 3-deoxy-n-mannooctulosonate (KDO), to a complex hydrophobic component, Lipid A. Anchored in the outer membrane, Lipid A is a derivative of D-glucosamine disaccharide and thus differs greatly in structure from the ubiquitous glycerophospholipids. Genetic blocks at different stages in the stepwise biosynthesis of core lipopolysaccharide can result in the conversion of the wild type (smooth) bacterial strains to rough or deep rough mutants which lack the O-antigen and a part or all of the core region.
Each of the three regions in LPS is endowed with multiple but specific biological functions (2, 3). The O-antigen, the determinant of the antigenic specificity of the organism, also functions as a receptor for one or more bacteriophages.
The core region, which can also function as a bacteriophage receptor, plays a role in maintaining the integrity of the outer membrane. Thus, in deep rough mutants which lack most of the core oligosaccharides, there is observed an increase in permeability to hydrophobic compounds concomitant with a decrease in protein content of the outer membrane (4-6). A similar effect is observed following the partial release of LPS from the membrane on EDTA treatment (7). Lipid A is the primary agent responsible for the endotoxicity of Gram-negative bacteria and displays mitogenic stimulation, complement activation and pyrogenic induction, in addition to a number of other biological functions (2). Lipid A as well as KDO appear to be indispensable to the Gram-negative bacteria. Thus, although mutants lacking the O-antigen and core region have readily been isolated, no bacterial mutant lacking Lipid A has yet been identified. The only mutants defective in the synthesis of KDO that have been observed are of the temperature-sensitive type (8-10). While the structures of the core and O-antigen polysaccharide regions have been elucidated in a number of bacterial strains, including Salmonella and E. coli (2), detailed structures of the Lipid A component have until recently remained unclear. Mainly through the efforts of Liideritz et al., Lipid A of Salmonella typhimurium is known to be composed of a Dglucosamine disaccharide backbone which carries up to 6 long chain fatty acyl residues, 2 phosphate residues, and, occasionally, 4-aminoarabinose and phosphorylethanolamine (11). Until recently, LPS was believed to be an oligomer consisting of D-glucosamine disaccharide subunits which were interlinked by means of phosphate groups. Both pyrophosphate (11) and phosphodiester bonds (12) have been postulated at one time Lipopolysaccharide from an E. coli Mutant 5907 or another to be present in Salmonella LPS. Very recently, from a "P NMR study (13), Salmonella LPS was concluded to be only a monomer of glucosamine disaccharide containing monoesterified phosphate groups, with the latter groups sometimes substituted to form phosphodiester or pyrophosphate linkages.
The work presented in this and the accompanying papers (14, 15) was undertaken with the principal aim of clarifying the structure of Lipid A. An E. coli K-12 deep rough mutant isolated and studied by Boman et al. (16) was chosen as the source of lipopolysaccharide.
It was hoped that this mutant, which contains only the KDO units as the carbohydrate component, would facilitate structural studies on Lipid A. In the present paper, we describe the isolation and characterization of a number of chemical degradation products derived from the LPS of the above strain. In the following paper (14), the chemical nature of the phosphate groups in LPS and its degradation products has been studied by 'IP NMR spectroscopy. In efforts directed toward the use of enzymes in structural elucidation, two long chain fatty acyl amidases were discovered. Their partial purification and use in structural work on Lipid A is described in the last paper of this series (15 In most experiments, 0.5% SDS was also added to the incubation mixture.
The action of the enzyme was restricted to the 4'-phosphate. Hydrolysis of the phosphomonoester group at the C1 position, which was usually slow, was completely inhibited by SDS.

Reduction with Sodium Borohydride
A IOO-~1 sample (20 nmol of glucosamine) was adjusted to pH 8 with 0.1 N NaOH, 100 ~1 of 1 M NaBH, was added, and the mixture was then incubated in a sealed screw capped vial at 52°C for 12 h. The sample was cooled on ice, acidified to about pH 2 with HCl, and the sample was dried in vacua by several evaporations of added methanol.

Treatment with Acetic Anhydride in Pyridine
A sample of "'P-labeled compound (10 to 100 nmol of phosphate) was rendered anhydrous by repeated evaporation of added dry pyridine. To this sample was added 100 ~1 of acetic anhydride and 100 ~1 of dry pyridine.
The reaction mixture was left at room temperature overnight and then dried down in vacua. To the residue was added 100 ~1 of aqueous pyridine for several hours. After evaporation, the residue was treated with 100 ~1 of 1 N NaOH at 0°C (15 min to 1 h), the solution neutralized with pyridinium Dowex 50 and lyophilixed. Aliquots of the reaction mixture were analyzed at different intervals by DEAE-cellulose thin layer chromatography in Solvent I (29). They were harvested and treated as described by Galanos et al. (33).
From Radioactively Labeled Cells-To radioactively labeled dried cells from a IO-ml culture was added 1 ml of phenol:chloroform: petroleum ether (2:5:8, v/v/v) solvent. After mixing with a glass rod, the mixture was centrifuged at 5,000 rpm for 15 min. The supernatant solution was combined with that obtained above from the unlabeled cells (0.5 g) and then treated as described above (33)

Purification and Characterization
of LPS Purification-LPS prepared by extraction with the organic solvent mixture was practically insoluble in water. Solubilization was accomplished either by warming or vigorous shaking in a TEAB solution (pH 7.5) or by treatment with EDTA (50 mM) in the presence of TEAB (pH 7.5) at 37°C for about 18 h followed by dialysis against water. Water-soluble preparations of LPS were thus obtained at concentrations up to 3 mg/ml.
The flow-through Bio-Gel (0.5 M) pattern of a preparation of LPS made using "*P-labeled cells gave a single sharp peak (Fig. 2, supplemental material in miniprint section).L However, tic on DEAE-cellulose in Solvent I gave two phosphoruscontaining products (Fig. 3) in about equal amounts. (In fact, as shown below, the 'lP-labeled LPS is a mixture of three components.) Composition-The composition of a preparation of LPS is shown in Table IA (miniprint  supplement).* In most analyses, the molar ratio of glucosamine:phosphate was close to 2:3; however, the ratio could also approach 212 depending upon the relative proportion of the individual LPS components in the particular LPS preparation (see below). Further, the molar ratio of KDO:glucosamine approached 3:2. Analyses of the fatty acid constituents of D31m4 Lipid A by gas-liquid chromatography indicated that lauric acid, myristic acid, and /?-hydroxymyristic acid were present in molar ratios approaching 1:1:3 ( However, no significant (~5% of total phosphorus) release of low molecular weight phosphate ester was observed following alkaline treatment of '"P-labeled LPS (Figs. 3 and 4). Similar treatment of 14C-labeled LPS released only [OH-LPS] and free fatty acids. Further, analyses by tic in Solvents I and IV of products formed by acidic treatment of '?-labeled [OH-LPS] showed that all products in the region where glucosamine and other nonancylated sugars might be expected to travel were KDO-positive.
Finally, no phosphate diester bonds were observed by 31P NMR in these LPS preparations (see following paper (14)).

Alkali-treated LPS
Purification-["2P]LPS was treated with alkali as described under "Methods." Although passage through a Bio-Gel 0.5m agarose column in 20 InM TEAB solution gave a single sharp peak of material included in the gel, tic in Solvents I and II showed separation into two major components (Fig. 4) Procedures," was applied dvectly to the column in a volume of 1 ml. Elution was with 0.02 M TEAB. Fractions of 1 ml volume each were collected at a flow rate of 0.3 ml/min and counted Cerenkov. The minor Peak 1 consisted of mixed micelles of partially degraded LPS and [OH-LPS] which eluted in the void volume. Pea/z 2, the major included peak, contained [OH-LPS]. Pea/z 3 contained primarily inorganic phosphate. Inset, autoradiogram of selected fractions from the 0.5 M agarose column. Thin layer chromatography was on DEAE-cellulose using Solvent I. Fractions from Peaks 1, 2, and 3 are shown in Samples 1, 2, and 3, respectively. separated by chromatography on a DEAE-cellulose column using a TEAB salt gradient (Fig. 5). Isolation of the components on a small scale was achieved by preparative tic on DEAE-cellulose using Solvent I.

Similar results were obtained with ['4C]glucose-labeled LPS. Upon alkali treatment, only two products, [OH-LPS]-I and [OH-LPS]-II,
and free fatty acids were observed by tic in Solvents I and II.
The above results were obtained by using unfractionated preparations of LPS as the starting material. When [32P]LPS fractions separated by tic as shown in Fig. 3A were studied, the component with higher RF in LPS, on alkaline treatment, gave only one product corresponding in mobility to [OH-LPS]-I (Fig. 3B, Sample 2). However, the slower component in LPS after similar treatment gave both [OH-LPS]-I and [OH-LPS]-II (Fig. 3B, Sample 3). Thus, the slower spot in LPS evidently contained at least two species. Since there was no interconversion of the different species under the conditions used (see below), the results showed that the 32P-labeled LPS was composed of at least three species.

Qualitative analyses of [OH-LPS]
were based upon colorimetric spray reagents. Both components, visualized by sulfuric acid charring, contained KDO and phosphate. No ninhydrinpositive material was observed.

Quantitative analyses of [OH-LPS]-I and [OH-LPS]-II showed that [OH-LPS]-I
contained phosphate and glucosamine in a molar ratio of 1:1,47% of the total phosphate being acid-labile.
[OH-LPS]-II contained phosphate and glucosamine in a molar ratio of 2.7:2; in this case, 64% of the total phosphate was acid-labile.   (14)), the lack of sensitivity of LPS to phosphomonoesterase appeared to be due to steric hindrance caused either by the fatty acyl groups directly or by the tendency of LPS to form aggregates. [OH-LPSJ

Phosphomonoesterase treatment of "*P-labeled [(SH-LPS]-I or 32P-labeled [OH-LPS]-II
generally released about 20 to 30% of the total phosphate as inorganic phosphate (Fig. 6 in [OH-LPS] were shown to be in a monoester linkage (see following paper (14)). The limited sensitivity of [OH-LPS] to phosphomonoesterase was probably due to steric hindrance caused by the neighboring fatty acyl groups or by the KDO residues.

Acidic Treatment of [OH-LPSJ-I and [OH-LPS]-II
Selective Cleavage of KLiO Units-The results of the treat-'ment of [@I-LPS] with sodium acetate (pH 4.5) at 1OO'C are given in Fig. 7, the total products formed being shown in the tic pattern in the inset. The major product as a function of time, designated Compound I, represented about 50% of the total 32P of the [OH-LPS].
As seen from the time course of the products formed during the acidic treatment, Compound I was very stable under these conditions. At least one transient intermediate which had lost one or more KDO units was also observed (see kinetics).
The separated [OH-LPS] components were also treated with acid as described above and the products formed as a function of time are shown in Fig. 8   were treated with 0.1 N HCl at 100°C for 15 min. As seen in Fig. 9, the major product in both cases had identical mobilities on tic in two solvents (see also Fig. 7). This product was subsequently identified as Compound II (see below). Thus, the two [OH-LPS] components (I and II) have the same structural backbone except for the acid-labile phosphate groups at the glycosyl bond (Fig. 1).

Isolation of Compound I
Compound I, the major product formed on treatment of [OH-LPS]-I at pH 4.5, was separated either by preparative tic on DEAE-cellulose (Fig. 7) or by a combination of gel filtration and chromatography on a DEAE-cellulose column. Fig. 10 shows the pattern obtained on a Sephadex LH-20 column in 0.1 M TEAB containing 40% methanol. If required, further purification of the acid hydrolysis products could be achieved by subsequent chromatography on DEAE-cellulose (Fig. 11). Compound I, thus obtained, was homogeneous when subjected to thin layer chromatography using four solvent systems (I, II, V, and VI).

Characterization of Compound I
The composition is shown in Table II. Thus, the molar ratio of glucosamine:phosphate is 1.O:O.g. KDO was absent and no reducing group was found. Compound I was degraded by three different methods. A summary of the methods and the structures of the products are given in Fig. 12

Hydrolysis
to Compound II-The acidic lability of Compound I was studied by using both "'P-and 14C-labeled preparations. The results are shown in Fig. 13 (miniprint supplement). Thus, there was a rapid release of inorganic phosphate corresponding to about 50% of the total 02P. The second major "*P-labeled product, Compound II, amounted to 38% of the initial "P. When "C-labeled Compound I was used, Compound II contained 75% of the 14C radioactivity. Thus, the ratio of 14C:82P in Compound II was 1.97:1.00, twice that of the starting material. The remaining products found were pre- ]glucose-labeled Compound I. Samples of "*P-labeled Compound I (90 nmol of phosphate) or '?-labeled Compound I (50 nmol of phosphate) were incubated in a IOO-~1 reaction mixture containing 0.5% SDS and the phosphatase. Thin layer chromatography was on DEAE-cellulose in Solvent I. Samples were quantitated by scintillation counting as previously described. Products are expressed as per cent total '? cpm or phosphate in the reaction mixture. 0, formation of 14C-labeled Compound III; 0, formation of "'P-labeled Compound III; A, formation of inorganic [32P]phosphate (PJ. [14C]glucose-labeled substrate in the presence of 0.5% SDS. As shown in Fig. 14 Mild acid treatment: Samples (6 nmol of phosphate) were treated with 30 ~1 of 0.1 N HCl at 100°C for 15 min. Thin layer chromatography was on DEAE-cellulose in Solvent I. Following autoradiography, the radioactive bands were excised and the radioactivity measured as described under "Experimental Procedures." Phosphatase treatment: Samples (7 nmol of phosphate) were incubated in 50+1 reaction mixtures in the presence of 0.5% SDS with alkaline phosphatase as described under "Experimental Procedures." Separation of products and their quantitation were as above. Compound I Degradation to Compound IV-'?-labeled Compound II, the acidic hydrolysis product of 14C-labeled Compound I, was treated with alkaline phosphatase. Over 90% of the "'C label was converted to a product that migrated near the front in Solvent System I, as expected for Compound IV (Table III). Similarly, inorganic phosphate was the sole "2P-containing compound observed following monophospha-tase degradation of "'P-labeled Compound II. Thus, the phosphate group in Compound II is monoesterified.
Treatment of Compound III with mild acid (0.1 N HCl, lOO'C, 15 min) as described above yielded the same product as that obtained by monophosphatase degradation of Compound II. Inorganic phosphate and Compound IV were the sole products of acidic hydrolysis of "'P-and ?Xabeled Compound III, respectively (Table III) 15. Autoradiogram of "P-labeled Compound I and inorganic [ '*P]pyrophosphate following degradation by oxidation and base-catalyzed elimination. Reaction conditions were as described under "Experimental Procedures." Thin layer chromatography was on DEAEcellulose in Solvent I. Sample 1, untreated "P-labeled Compound I; Sample 2, untreated inorganic ["'Plpyrophosphate; Sample 3, '*Plabeled Compound I following oxidation/P elimination treatment; Sample 4, inorganic ["2P]pyrophosphate following oxidation/p elimination treatment.
that the monophosphatase-sensitive phosphate residue and the glycosidically linked acid-labile phosphate group are different and complementary in their sensitivity to acid and to phosphatase.
Composition of Compounds II to IV-As seen in Table II, and as expected, Compounds II and III both had a glucosamine:phosphate ratio of 2:0.9, signifying loss of 50% of the phosphate in Compound I.
Compound IV was reduced with sodium borohydride and then subjected to hydrolysis by acid. Subsequent analysis of reducing glucosamine (Table II) showed that of the 2 molar equivalents of glucosamine in Compound IV, 0.7 mol was nom-educing.
The result was consistent with a disaccharide structure.

Degradation by Oxidation Followed by ,8 Elimination-
Alkali-catalyzed elimination of substituents in a position ,6 to electron-withdrawing centers is a classical method for degradation of long polysaccharide chains (35). This technique was utilized as a means of exploring both the nature and points of linkages of the phosphate residues. 32P-labeled Compound I was oxidized with acetic anhydride in dimethyl sulfoxide as described under "Experimental Procedures." Following basecatalyzed elimination, the products were examined by high voltage paper electrophoresis and thin layer chromatography in Solvent I. Inorganic phosphate was formed and its yield was as high as 63% of the total phosphate (Fig. 15). Similarly, following oxidation of Compound II and subsequent p elimination at pH 9, 20% of the total phosphate was released as inorganic phosphate.
No inorganic pyrophosphate was detected. A sample of inorganic pyrophosphate treated under similar conditions was recovered unchanged (Fig. 15); therefore, it is unlikely that inorganic pyrophosphate was released and subsequently degraded during the oxidation treatments. These results suggest that Compound I contains two phosphomonoester groups in a position p to oxidizable hydroxylic groups in the sugar residues. This is consistent with one phosphate group at C-4 of the nonreducing glucosamine residue and one phosphate in a glycosidic linkage, in analogy to the structure concluded for Salmonella Lipid A (13). A phosphate group at C-6 of the nonreducing glucosamine residue would also yield similar results; however, this possibility is eliminated on the basis of 'IP NMR studies (see following paper (14)). The degradation scheme is shown in Fig. 16. The above uncertainties in the use of the alkaline phosphatase invalidated the application of another chemical method for the characterization of diesters of pyrophosphoric acid. The latter, on treatment with acetic anhydride/pyridine at room temperature, are cleaved to form phosphomonoesters (41). The mild method is thus diagnostic of diesters of pyrophosphates.
However, the method was inapplicable in the present work because of the unreliability of the assay involving the phosphomonoesterase.
Finally, the influence of the medium used for growth of E. coli cells on the composition and structure of LPS deserves comment. Cells grown in low phosphate media revealed two fractions of [LPS]-I which appeared to differ in the extent of esterification by the fatty acyl groups. Two fractions of [LPS]-II, which probably differ in fatty acid content as well, were identified by "'P NMR studies of LPS isolated from cells grown in rich media (14). Variation in the ratio of [LPS]-I: [LPS]-II from about 2:l for cells grown in low phosphate media to about 1:2 for cells grown in rich media was also observed. Thus, growth conditions clearly influence the resultant composition of the lipopolysaccharide (and Lipid A) and it is possible that there are other derivatives of LPS in